GitLab 19.3 Bolsters Enterprise Control Over Agentic Development
GitLab 19.3 enhances enterprise security and control for agentic software development, allowing dedicated customers to run AI platforms within their single-tenant environments and manage AI data securely.
GitLab's latest release, version 19.3, introduces significant updates aimed at providing enterprises with greater control and security as they adopt agentic software development practices. A key highlight is the extension of the GitLab Duo Agent Platform into GitLab Dedicated's single-tenant environments. This allows customers to run these sensitive AI workloads within their existing secure infrastructure, connect their own AI models for inference, and crucially, keep AI-processed data within their defined security boundaries.
The update also brings enhanced support for critical security features such as Secrets Manager, Flow Creator Agent, and advanced SAST capabilities including Bulk SAST False Positive Detection and Agentic SAST Vulnerability Resolution. These additions aim to streamline development workflows while maintaining robust enterprise-grade control over AI-driven processes. "Every capability we shipped this month, from where an agent runs to which secret it can touch, extends that same control into the trusted software delivery workflows enterprises already depend on," stated Manav Khurana, chief product and marketing officer at GitLab.
The AI Gateway for the GitLab Duo Agent Platform now operates within GitLab Dedicated's single-tenant SaaS infrastructure. This ensures that agentic workloads adhere to the same residency and isolation models as other software development lifecycle components. This integration is particularly beneficial for regulated industries and teams with strict data residency requirements, enabling them to leverage agentic AI under familiar and auditor-approved deployment models.
GitLab Secrets Manager, currently in limited availability as a paid add-on, extends control over secrets beyond the traditional CI pipeline. It now supports a wider range of tools including Kubernetes, Terraform, OpenTofu, and custom solutions. By centralizing credentials within the same platform that manages code and pipelines, teams can eliminate the need for separate permission models, thereby reducing complexity and potential security gaps.
The Bulk SAST remediation features are designed to tackle accumulated technical debt efficiently. Teams can now select multiple SAST findings within the Vulnerability Report, receive a confidence score for each, and generate ready-to-merge fixes. This capability allows developers to review and merge fixes rather than writing them from scratch, significantly accelerating the remediation of security vulnerabilities across large codebases.
For automation enthusiasts, the Flow Creator Agent removes the complexities of manual schema mapping. Available through Agentic Chat, this foundational agent allows users to describe desired automations in plain language, generating complete and runnable flows. These flows are then registered via the AI Catalog, with each still operating under a scoped service account and requiring appropriate permissions for enablement.
Additional enhancements in GitLab 19.3 include the general availability of GitLab Credits usage caps, enabling organizations to set monthly spending limits on agentic AI services. Furthermore, restricted visibility for custom agents and flows per GitLab group is now generally available, enhancing access control and collaboration across multiple projects within a group.